Avoca built an AI back office for HVAC, plumbing, and home trades – then went quiet for two years before emerging with 800 clients and a $1B valuation.
ENTRY ANGLES
AI platforms for quality inspection in physical infrastructure (cell towers, power lines) · AI automation for unglamorous B2B markets overlooked by tech founders · Replacing manual processes (printed instruction binders) with software in manufacturing
VERTICALS
CAPABILITIES
AI/computer vision for quality inspection, Domain expertise in unglamorous B2B markets, Ability to replace analog workflows with software solutions
AVOCA FOUNDER
“Home services is not a particularly glamorous category in Silicon Valley,”
This startup graduated from Y Combinator in spring 2023 and went quiet. Two days ago it came back – announcing multiple funding rounds totaling $125M, with its most recent valuation hitting $1 billion. That kind of number was impossible to scroll past.
Avoca built an AI back office for companies in the home services trades: HVAC installation and repair, plumbing, electrical, roofing, pest control, and related fields.
Avoca's back office covers four functional areas. On the customer-facing side, AI agents handle all inbound communication – answering calls, responding to emails, replying to messages across chat platforms, clarifying the problem, proposing solutions, and scheduling technician visits. Requests too complex for the AI are routed to a human dispatcher.
That communication layer integrates directly with a scheduling system that coordinates technician availability and syncs appointment times with customers.
Sitting underneath both is paid advertising: campaign creation, deployment, monitoring, and optimization run automatically to generate the customer inquiries that the communication layer then processes.
Completing the picture is an AI coach that analyzes all customer interactions – before, during, and after a job – surfacing insights about why customers drop off before booking, what causes dissatisfaction, and how communication quality affects close rates. Companies use this to improve interaction standards and optimize pricing, turnaround times, and quality benchmarks.
Avoca's pivot story is worth telling. The founders graduated from Y Combinator with an AI platform that handled inbound calls for restaurants and were actively selling it. Then, at a restaurant technology conference, they met executives from an HVAC company who were also interested in the platform.
The insight hit immediately. When a restaurant misses a call, it loses $30–40 on an order. When an HVAC company misses a call, it potentially loses $30,000–40,000 on a whole-home system installation. Same concept, same technology – entirely different order of magnitude in problem severity.
The founders spent three months embedded with that HVAC company, adapting the platform for the new vertical. When the rebuilt product was working, that first customer connected them to others in the space. The founders started attending home services trade shows, which yielded their first 10 customers. From there, gradual growth brought them to 800 customers today – who will collectively route more than $1 billion in job orders through Avoca's platform this year.
"Home services is not a particularly glamorous category in Silicon Valley," said a Kleiner Perkins partner who led the investment in Avoca. "The reason is that most tech founders have no idea how large and important this market actually is"
The US HVAC market alone was $50B in 2025 and is projected to grow to $75B by 2032. And the key constraint in this market is physical labor. Robots won't enter these industries anytime soon.
This is a consequence of what's called Moravec's paradox: it turns out to be far easier to automate high-level cognitive tasks (chess, calculus) than low-level sensorimotor skills (the kind of perception and fine motor control that a one-year-old manages effortlessly). The work of skilled tradespeople falls squarely in the latter category – and it happens in endlessly varied, unpredictable environments, not on a controlled factory floor.
So AI enters these markets not by replacing workers, but from the operational side – AI platforms that coordinate, optimize, and improve the efficiency of human labor.
That said, the claim that Silicon Valley "completely ignores" these markets overstates things. Relative to the market's scale, the number of startups attacking it remains small – but they exist.
Cactus ([covered here](/review/sjuda-eshhjo-mozhno-uspet)) – another Y Combinator alumni – traveled a similar path to Avoca. It started with an AI platform for scheduling calls for various self-employed service providers before pivoting to AI software specifically for home services companies, raising $7M in November.
Elyos ([covered here](/review/sotni-tysjach-kompanij-u-kotoryh-ii-ne-otberjot-rabotu)), a British startup also from Y Combinator, builds an analogous platform and raised £9.7M (roughly $13M) in January.
Siro ([covered here](/review/smotri-ka-ved-takie-prodazhi-tozhe-nuzhno-uluchshat)) made a savvy pivot – from AI coaching for phone sales to coaching for in-person home services visits – and immediately raised $50M. Its AI coach runs in the background during technician conversations with customers and provides post-visit feedback on what could be improved.
Robby ([covered here](/review/prostye-veshhi-ogromnyj-rynok-bolshie-dengi)) – a current Y Combinator batch member – is building a similar AI coaching product for the same home services vertical.
It's becoming clear that Silicon Valley doesn't just overlook one large unglamorous market – it overlooks several.
A founder of Squint ([covered here](/review/zdes-u-tebja-100-shansov-pobedit-glavnogo-konkurenta)) made the same observation about industrial manufacturing: "For some reason, manufacturing is an abandoned topic in the tech world. Even now, it's nearly impossible to find founders building in manufacturing in Silicon Valley. So when we talk about our main competitors, we're not talking about other startups – we're talking about binders full of printed instructions."
That quote came to mind a few days ago reviewing Cloneable ([covered here](/review/mnogo-deneg-i-malo-konkurentov)), which raised $4.6M for an AI platform automating quality inspection of cell towers and power line installations.
The pattern holds: build AI platforms for markets that tech founders don't look at because they seem insufficiently exciting. Home services, manufacturing, physical infrastructure maintenance, and adjacent fields.
The question is simply which of these markets you have the courage to look at – knowing that the courage pays off handsomely, as Avoca's billion-dollar outcome makes clear.